If robots learn from people, people should share in the value they create.

The right to know exactly what you’re being asked to teach.

Before recording begins, Teachers should understand the task, the expected duration, the capture requirements, the equipment, the quality standards, the acceptance criteria and how they will be paid.

No mystery jobs.

The reward and requirements a Teacher accepts should be the ones they are held to. Anything that can change later — like royalty rates for future licenses — should say so up front, and a change should never reach back into what a Teacher has already earned.

The right to understand the license.

A Teacher should not need a law degree to understand what happens to their work.

Before submitting a Lesson, they should be told in plain language:

who owns the recording, what rights they are granting, whether it can be licensed more than once, whether it is exclusive, what uses are permitted, and what they can and cannot take back later.

A buried clause should never be the business model.

The right to know when their work is licensed, and for what.

Every time a Teacher’s work is licensed, they should be able to see that it happened, when, and the general purpose it was licensed for.

A customer’s name may legitimately be kept confidential. The licensing event itself never should be.

Training data shouldn’t disappear into a black box the moment it leaves the Teacher’s hands.

The right to participate economically when their work creates value.

Human demonstrations can keep creating value long after they are recorded.

When a Teacher’s work generates licensing revenue, the Teacher should share in that revenue under published terms they can read before they submit.

The human should not be the only participant in the value chain with zero downstream upside.

The right to participate again when reusable work is licensed again.

If the same non-exclusive Lesson creates value for another customer, there should be another economic event for the Teacher.

Licensed once? Earn once.

Licensed again? Participate again.

Reusable data should have reusable economics.

The right to see the math.

When a Lesson is licensed, Teachers should be able to see how their earnings were calculated: the license value attributed to their work, their share, what it earned them, and whether it has been paid or reversed.

A royalty you cannot check isn’t much of a royalty.

The right to retain ownership of their original contribution.

Submitting a Lesson should not silently mean surrendering ownership of it forever.

Teachers should keep ownership of their original recordings unless they knowingly agree to a different arrangement.

The platform and its customers receive the licenses that were actually agreed to. Nothing more.

The right to choose exclusivity.

Exclusive data can be valuable. It also removes the Teacher’s ability to earn from additional customers.

Any exclusive arrangement should therefore be clearly identified before the Teacher records, and paid under separately disclosed terms.

Exclusivity should be a choice, not a surprise.

The right to stop future licensing.

Teachers should be able to remove eligible Lessons from future licensing.

That does not rewrite history. Licenses already granted may remain valid, and models already trained on properly licensed data may not be practically reversible.

But removing a Lesson should stop new licenses from being issued for it going forward.

Control should be real, even when it cannot be retroactive.

The right to privacy, dignity and safe participation.

Teaching robots should never require someone to unnecessarily surrender their privacy or safety.

Teachers should receive clear guidance about sensitive information, private spaces, faces, voices, children, bystanders, confidential materials and anything else that should not be captured.

Teachers should be free to decline or walk away from any Lesson that feels unsafe, invasive, illegal or inappropriate — without penalty.

You should not have to trade dignity for access to work.

The right to provenance.

Training data should carry a record of where it came from: how it was captured, and the consents and permissions behind it, as it moves through the AI supply chain.

The Teacher’s personal identity does not need to travel everywhere their data does.

Their rights do.

The right to a fair review and recourse.

If work is not accepted, Teachers should be told why.

If they believe a mistake was made, they should have a real way to raise it and ask questions.

And work that is turned away should never be licensed or sold anyway. If a platform decides a Lesson does not qualify for its marketplace, that Lesson should not later turn up in a dataset sold to its customers.